Resumen de Computer's imagination creates human faces
Alec Radford of machine learning firm Indico in Boston and Facebook's Al research lab in New York have looked at a particular type of network called a generative adversarial network. In these, one part of the system tries to invent fake data to tool another part into mistaking it for training data. The idea is that by repeatedly pitting the network against itself, it will learn to produce better results. In one experiment, the team trained the network on pictures of faces and then tried out a kind of visual arithmetic. They gave the network pictures of smiling women, then told it to "subtract" pictures of women with neutral expressions and "add" men with neutral expressions. The goal was to extract the concept of "smiling" and combine it with the concept of "man." The results were entirely imagined pictures of smiling men
